Behavioral/Cognitive

Neurons in the Primate Medial Basal Forebrain SignalCombined Information about Reward Uncertainty, Value,and Punishment Anticipation

Ilya E. Monosov,1,3 X David A. Leopold,2,4 and X Okihide Hikosaka3

1Department of Anatomy and Neurobiology, Washington University, School of Medicine, St. Louis, Missouri 63110, 2Section on Cognitive Neurophysiologyand Imaging, National Institute of Mental Health, Bethesda, Maryland 20892, 3Laboratory of Sensorimotor Research, National Eye Institute, Bethesda,Maryland, 20892, and 4Neurophysiology Imaging Facility, National Institute of Mental Health, National Institute of Neurological Disorders and Stroke,National Eye Institute, Bethesda, Maryland 20892

It has been suggested that the basal forebrain (BF) exerts strong influences on the formation of memory and behavior. However, whatinformation is used for the memory-behavior formation is unclear. We found that a population of neurons in the medial BF (medialseptum and diagonal band of Broca) of macaque monkeys encodes a unique combination of information: reward uncertainty, expectedreward value, anticipation of punishment, and unexpected reward and punishment. The results were obtained while the monkeys wereexpecting (often with uncertainty) a rewarding or punishing outcome during a Pavlovian procedure, or unexpectedly received an out-come outside the procedure. In vivo anterograde tracing using manganese-enhanced MRI suggested that the major recipient of thesesignals is the intermediate hippocampal formation. Based on these findings, we hypothesize that the medial BF identifies various contextsand outcomes that are critical for memory processing in the hippocampal formation.

Key words: basal forebrain; electrophysiology; emotion; learning; memory; motivation

IntroductionThe basal forebrain (BF) is thought to be a key brain area for thecontrol of cognitive functions, such as learning and memory(Mesulam et al., 1983; Everitt et al., 1988; Voytko, 1996; Everittand Robbins, 1997; Semba, 2000; Zaborszky et al., 2015). Lesionsof BF in humans produce deficits in new memory formation andrecall (Damasio et al., 1985; Morris et al., 1992; Abe et al., 1998;Solca et al., 2015), and the severity of mnemonic symptoms inAlzheimer’s patients is often related to BF degeneration (White-house et al., 1982; Kesner, 1988).

Animal studies further support the wide role of the BF in thecontrol of computational resources for a range of cognitive func-tions, such as perception, attention, learning, and memory (Berryand Thompson, 1979; Ridley et al., 1989; Muir et al., 1993; Has-selmo et al., 1996; Voytko, 1996; Everitt and Robbins, 1997; Rid-ley et al., 1999a, b; Savage et al., 2007; Baxter and Bucci, 2013) andsingle out the medial regions of the BF (mBF) as particularly

important for many kinds of new learning and long-term mem-ory formation. mBF lesions have been found to impede classicalaversive learning (Berry and Thompson, 1979) and have alsobeen shown to influence associative memory formation (Has-selmo et al., 1996; Everitt and Robbins, 1997). However, whattypes of information mBF neurons send to its target structures tomodulate learning and memory is unknown.

Learning and memory are known to be disproportionallyshaped by surprising events or uncertain contexts (Pearce andHall, 1980; Yu and Dayan, 2005; Courville et al., 2006; Esber andHaselgrove, 2011; Le Pelley et al., 2011; Bach and Dolan, 2012;Ogawa et al., 2013), as well as by other factors, such as the expec-tation of a reward or the fear of an aversive stimulus (Ecclestonand Crombez, 1999; Bromberg-Martin et al., 2010; McNally etal., 2011; Moore et al., 2012; Anderson, 2013; Wiech and Tracey,2013). We therefore recorded activity of single mBF neuronswhile monkeys experienced contexts known to activate and mod-ulate a wide range of learning and memory functions. We foundthat a population of neurons in the mBF combined informationabout uncertainty, reward size, punishment, and surprise. Wealso found that the region of the mBF where these neurons wereclustered projects to the hippocampal formation (HF). Based onthese findings, we propose a mechanism through which the mBFmay control learning and memory functions of the HF.

Materials and MethodsGeneral procedures. Five adult male rhesus monkeys (Macaca mulatta)were used for the experiments (Monkeys H, P, T, Sm, and S). All proce-dures for animal care and experimentation were approved by the AnimalCare and Use Committee of the National Eye Institute and complied with

Received Jan. 5, 2015; revised March 9, 2015; accepted March 31, 2015.Author contributions: I.E.M. designed research; I.E.M. performed research; I.E.M. analyzed data; I.E.M., D.A.L.,

and O.H. wrote the paper.This work was supported by the National Eye Institute intramural research program and the Department of

Anatomy and Neurobiology, Washington University School of Medicine to I.E.M. We thank E. Bromberg-Martin, B.Cumming, P. Daye, A. Ghazizadeh, J. Herman, H. Kim, R. Krauzlis, L. Optican, R. Wurtz, and M. Yasuda for valuablescientific discussions; F. Ye and C. Zhu for excellent MRI services; M. Smith for histological expertise and service; andA. Hays, J. McClurkin, B. Nagy, N. Nichols, D. Parker, and T. Ruffner for technical support.

The authors declare no competing financial interests.Correspondence should be addressed to Dr. Ilya E. Monosov, Department of Anatomy and Neurobiology, Wash-

ington University, School of Medicine, 660 S. Euclid Avenue, St. Louis, MO 63110. E-mail: ilya.monosov@gmail.com.DOI:10.1523/JNEUROSCI.0051-15.2015

Copyright © 2015 the authors 0270-6474/15/357443-17$15.00/0

The Journal of Neuroscience, May 13, 2015 • 35(19):7443–7459 • 7443

the Public Health Service Policy on the humane care and use of labora-tory animals. A plastic head holder and plastic recording chamber werefixed to the skull under general anesthesia and sterile surgical conditions.The chambers were tilted laterally by 35° and aimed at the ventromedialprefrontal cortex and the anterior portion of the caudate nucleus. Twosearch coils were surgically placed under the conjunctiva of the eyes.After the monkeys recovered from surgery, they were conditioned usingPavlovian procedures (Experiments 1–5). During the Pavlovian proce-dures, we recorded the activity of single neurons in the basal forebrain–septum complex.

Neuronal recording. While the monkey was participating in the task, werecorded the activity of single neurons in the septum and the diagonalband of Broca. The recording sites were determined with 1 mm spacinggrid system, with the aid of MR images (4.7T, Bruker) obtained along thedirection of the recording chamber. This MRI-based estimation of neu-ron recording locations was aided by custom software (Daye et al., 2013).

Single-unit recording was performed using epoxy-coated electrodes(FHC). The electrode was inserted into the brain through a stainless-steelguide tube and advanced by an oil-driven micromanipulator (MO-97A,Narishige). The electric signal from the electrode was amplified with abandpass filter (200 Hz-10 kHz; BAK). Neuronal spikes were isolatedon-line using a custom voltage-time window discrimination software(MEX, LSR/NEI/NIH).

Histological procedure. After the end of some recording sessions, wemade electrolytic microlesions at the recording sites (15 �A and 30 s) inwhich we encountered typical Type 2 neurons in the mBF. After theconclusion of experiments, the monkey was deeply anesthetized usingsodium pentobarbital and perfused with 10% formaldehyde. The brainwas blocked and equilibrated with 10% sucrose. Frozen sections were cutevery 50 �m in the coronal plane. The sections were stained withcresyl-violet.

Data processing and statistics. Spike-density functions were generatedby convolving spike-times with a Gaussian filter (� � 50 ms). The con-ditioned stimulus (CS) neuronal responses were measured during a win-dow of 150 ms after CS presentation until the end of the CS epoch. Tonormalize task-event related responses (Experiments 2–5), we subtractedbaseline activity (last second of the intertrial interval) from the activity

during the task-event related measurement epoch. All statistical testswere two-tailed. For comparisons between two task conditions for eachneuron, we used rank-sum test, unless otherwise noted. For comparisonsbetween two task conditions across the population average we used apaired signed-rank test, unless otherwise noted.

Experiment 1: appetitive and aversive Pavlovian procedure. For this ex-periment, we recorded widely in the septal-basal forebrain complex inMonkeys P and Sm. To confirm our results in Monkeys P and Sm, weincluded data from two additional monkeys (T and S) used in a previousstudy (Monosov and Hikosaka, 2013).

The procedure consisted of two alternating blocks: appetitive andaversive blocks (Fig. 1). In the appetitive block, each of three CSs werefollowed by an unconditioned stimulus (US) (apple juice), with 100%,50%, and 0% chance, respectively. In the aversive block, three CSs werefollowed by an airpuff US directed at the monkey’s face with 100%, 50%,and 0% chance, respectively. The 0.4 ml of apple juice was deliveredthrough a spout that was positioned in front of the monkey’s mouth.Airpuff (�35 psi) was delivered through a narrow tube placed 6 – 8 cmfrom the monkey’s face.

Each trial started with the presentation of a trial-start cue (whitecircle) at the center of the screen. After 1 s, the trial-start cue disap-peared and one of the three CSs was presented pseudorandomly at thecenter. After 1.5 s, the CS disappeared, and the US (if scheduled forthat trial) was delivered. The monkeys were not required to fixate thetrial-start cue or CS. For Monkeys P and Sm, one block consisted of12 trials with fixed proportions of trial types (100%, 4 trials; 50%, 4trials; 0%, 4 trials), and the intertrial intervals (ITIs) ranged from 3 to8 s. For Monkeys S and T, one block consisted of 22 trials, and the ITIsranged from 5 to 10 s.

Neuron’s uncertainty-preference was defined in Experiment 1 if itsresponses varied across the three CSs in either appetitive or aversive block(Kruskal–Wallis test, p � 0.05) and if its response to the uncertain CS(50%) was significantly stronger than its responses to both of the certainCSs (100% and 0%) ( post hoc test: two-tailed rank-sum test; p � 0.01).To assess the heterogeneity of uncertainty sensitive neurons, we per-formed principal component analysis Figure 2). To quantify whether aneuron had a significant airpuff response, we compared the activity in a

Figure 1. Responses of Type 2 neurons in the mBF to predictions about rewards and punishments. A, Monkeys experienced two distinct blocks: an appetitive block in which three visual CSspredicted juice (US) with 100%, 50%, and 0% chance, and an aversive block in which three cues predicted airpuff (US) with 100%, 50%, and 0% chance. TS, Trial start cue. B, Responses of a singleType 2 neuron in two blocks. Its spike activity is shown by raster plots (top) and spike density function (bottom). Dark blue raster plots indicate the activity in 50% CS trials in which US was omitted.

7444 • J. Neurosci., May 13, 2015 • 35(19):7443–7459 Monosov et al. • Neurons in the Primate Medial Basal Forebrain

50 ms time window starting from 20 ms after airpuff delivery (or omis-sion) with the activity in a 50 ms time window before airpuff delivery oromission.

For each neuron, the presence of ramping (Fig. 3D) was assessed witha correlation of neuronal activity (sampled every 100 ms) and time dur-ing the last 1 s of the CS period. Ramping activity was considered presentif the correlation was positive and significant. The statistical significanceof the correlation ( p � 0.05) was tested using a permutation test (nullhypothesis: neuronal activity was independent of time) by shuffling theneural firing rates across the time bins 2000 times. The slope of theanticipatory ramping activity was defined as the slope of the regressionline fit to these data.

The intrinsic firing characteristics of Type 2 neurons were defined asfiring irregularity, autocorrelation width, and firing rate (Fig. 4). Toquantify irregularity of spiking activity, we used the irregularity index(IR) developed by Davies et al. (Davies et al., 2006; Nakamura et al., 2008;Matsumoto and Hikosaka, 2009).This measurement does not requireconstant firing during the measurement period and is therefore useful foranalyzing neurons with different firing properties. To verify our IR-basedresult, we also quantified spike rate variance, defined as the SD of inter-trial intervals (ISIs) during the same baseline window. Neuronal bursti-ness was assessed during the same baseline window for each neuron usingtwo measures previously used to study hippocampal burst neurons (Vis-kontas et al., 2007): (1) the width of spike autocorrelation (computed in

1 ms bins) defined as the first time (lag) the autocorrelation reached itsmean point; and (2) the ratio of the number of ISIs �10 ms and thenumber of ISIs �10 ms.

Experiment 2: unexpected reward and punishment delivery. The proce-dure was the same as in Experiment 1, except that uncued trials wereincluded in which a juice reward alone or an airpuff alone was deliv-ered unexpectedly during some of the ITIs. These unexpected out-comes were included during 20% of the ITIs, which were chosenrandomly. When the outcome was delivered, it occurred between�1.3 and 3 s after the end of the previous trial. Monkeys H and Pparticipated in this experiment.

To compare the average reward or punishment responses after 50%CSs with average responses for the same outcomes delivered during theITI (Fig. 5C,E), a paired signed rank test was performed comparing theactivity in a 150 ms window after outcome delivery. To calculate anddisplay the outcome response after the 50% CSs, the activity during 50%trials in which the outcome was not delivered was subtracted from 50%trials in which the outcome was delivered. To calculate and display theoutcome response activity during the ITI, average baseline wassubtracted.

Experiment 3: reward probability and reward amount Pavlovian proce-dure. The procedure consisted of two blocks: a reward probability blockand a reward amount block (Fig. 6). Monkeys H and P participated inthis experiment.

Figure 2. Clustering of two groups of reward uncertainty-sensitive neurons. A, The CS epoch activity for 70 uncertainty selective neurons during the appetitive/aversive procedure. Each tracerepresents the average activity of a single neuron during the 6 CS epochs. We normalized each trace by dividing each of these 6 CS responses by the maximum CS response. PCA was performed onthose normalized response functions. Orange and red traces represent the two clusters of neurons (Type 1 and Type 2) that resulted from the PCA analysis (shown in B, C). B, PCA performed on thesingle neuron functions in A revealed that the majority of the variance in the 70 neurons’ CS activations can be explained by the first principal component (PC1). To assess this, we shuffled normalizedactivity across the 0% juice, 100% juice, 0% airpuff, 50% airpuff, and 100% airpuff CSs and calculated the percentage of variance explained (PVE) by each PC. Gray error bars represent the range ofPVE for each PC based on 10,000 shuffles. C, Bimodal distribution of PC1s of the 70 uncertainty neurons (Hartigan’s Dip Test, p � 0.01, tested by 10,000 permutations). Based on the PC1s, k-meansclustering separated the neurons into two groups: orange represents Type 1; red represents Type 2. Their average firing is shown in D for the three CSs in each block (appetitive or aversive), separately.

Monosov et al. • Neurons in the Primate Medial Basal Forebrain J. Neurosci., May 13, 2015 • 35(19):7443–7459 • 7445

In the reward probability block, five CSs were followed by a liquidreward (0.4 ml of juice) with 100%, 75%, 50%, 25%, and 0% chance,respectively. In the reward amount block, five CSs were followed by aliquid reward of 0.4, 0.3, 0.2, 0.1, and 0 ml, respectively. Thus, the ex-pected values of the five CSs matched between the probability andamount blocks. Each trial started with the presentation of a trial-start cueat the center. The trial-start cue was a purple square in the probabilityblock and a yellow square in the amount block. The monkeys had tomaintain fixation on the trial-start cue for 1 s; then the trial-start cuedisappeared and one of the five CSs was presented pseudorandomly.After 1.5 s, the CS disappeared, and juice (if scheduled for that trial) wasdelivered. The monkeys were not required to fixate the CSs. In each trial,the CS could appear in three locations: 10 degrees to the left or to the rightof the trial-start cue, or in the center. One block consisted of 40 trials withfixed proportions of trial types (each of the five CSs appears eight timeseach block).

In a separate experimental session, we tested the monkeys’ choicepreference for the CSs (Fig. 7). The preference was tested in three ways:among the probabilistic CSs, among the amount CSs, and across theprobabilistic and amount CSs. Monkeys H and P participated in thisexperiment. Each trial started with the presentation of the trial-start cueat the center, and the monkeys had to fixate it for 1 s. Then two CSsappeared,10 degrees to the left and right. The monkey had to make asaccade to one of the two CSs within 5 s and fixate it for at least 800 ms.Then the unchosen CS disappeared, and after �750 ms the outcome(associated with the chosen CS) was delivered and the chosen CS disap-peared. If the monkey failed to fixate one of the CSs, the trial was aborted

and all stimuli disappeared. The trials were presented pseudorandomly,so that a block of 180 trials contained all possible combinations of CSsfour times.

Experiment 4: reward variance Pavlovian procedure. The procedureconsisted of one block in which four fractals were presented as CSs (Fig.8). Monkeys H and P participated in this procedure. The “no variance”CS was always followed by 0.15 ml of juice. The “low variance” CS wasfollowed either by 0.2 or 0.1 ml of juice, with a 50% chance. The “me-dium variance” CS was followed either by 0.25 or 0.05 ml of juice, with a50% chance. The “high variance” CS was followed either by 0.3 ml ofjuice or no juice, with a 50% chance. Thus, the four CSs had the sameexpected value but different reward variances.

Each trial started with the presentation of a trial-start cue (gray square)at the center. The monkeys had to maintain fixation on the trial-start cuefor 1 s; then the trial-start cue disappeared and one of the four CSs waspresented at the center. After 1.5 s, the CS disappeared, and juice (ifscheduled for that trial) was delivered.

In a separate experimental session, we tested the monkeys’ choicepreference for the CSs (Fig. 9). Monkeys H and P participated in this task.The details are the same as in the choice procedure used in Experiment 3(above).

Experiment 5: reward probability Pavlovian procedure with novel stim-uli. The procedure was the same as the appetitive block in Experiment 1,except that novel fractals were used for the CSs (Fig. 10). Also, here themonkey (P) was expected to fixate the trial-start cue for 1 s (as in Exper-iments 1– 4). The monkey started learning some sets of CSs (three novelfractals for each set) in the Pavlovian procedure on consecutive days, one

Figure 3. Locations and firing properties of two groups of reward uncertainty-sensitive neurons in the septal area. A, B, Their locations are shown on MR images: left, coronal; right, parasagittal.Yellow represents Type 1 neurons in the anterodorsal septal region (n � 31). Red represents Type 2 neurons in the mBF (n � 39). Light green represents other recorded neurons (n � 300). C,Electrolytic marking lesions (black arrows) made at the locations of Type 2 neurons in the mBF of Monkey H. They were made along two mediolaterally adjacent electrode tracks. Along the lateraltrack, we recorded four Type 2 neurons between the two marking lesions. D, A table of response properties of Type 1 and 2 neurons recorded during Experiment 1. Asterisks indicate significantdifference between Type 1 and Type 2 neurons (baseline firing rate and irregularity index; Wilcoxon rank sum test, p � 0.05; proportions of neurons displaying significant CSs ramping or airpuffresponses; see Materials and Methods; � 2 test, p � 0.05). A–C, Gray lines indicate 5 mm scale bars. A, C, The coronal sections are located �1.5 mm anterior to the center of the anterior commissure(ac). cc, Corpus callosum; CD, caudate nucleus; DBB, diagonal band of Broca; lv, lateral ventricle; LS, lateral septum; MS, medial septum; PUT, putamen; vmPFC, ventromedial prefrontal cortex.

7446 • J. Neurosci., May 13, 2015 • 35(19):7443–7459 Monosov et al. • Neurons in the Primate Medial Basal Forebrain

session per day. On the second and third days, more sets of CSs wereadded each day for the monkey’s additional learning. Neuronal record-ing was done on the third day while the monkey was performing thePavlovian procedure with these sets of CSs. In other words, the mon-key had experienced the first group of CS sets for 2 d, the secondgroup for 1 d, and the third group was novel. In addition, a well-learned set of CSs was used during the neuronal recording. Thisacross-day procedure was done several times by introducing morenovel fractals as CSs.

Choice trials (as a block of nine trials) were regularly inserted to thePavlovian procedure (Fig. 11). On each trial, two of the three CSs werechosen and presented on the right and left (see Experiment 3). The pur-pose of the choice trials was to examine the changes in the monkey’sbehavior: choice preference, the number of saccades, and the time beforemaking the choice.

Experiment 6: manganese enhanced MRI tracing of mBF projections. Totest where the mBF Type 2 hotspot most strongly projects, we usedmanganese-enhanced MRI tracing (MEMRI) in Monkey P (Fig. 12). Themethod relies on two properties of manganese ions: (1) the manganese

ion (Mn 2�) is a calcium ion analog and is therefore taken up by neuronsand transported in an anterograde fashion; and (2) Mn 2� increases theMR intensity of voxels (Pautler et al., 1998; Saleem et al., 2002; Simmonset al., 2008). Therefore, by injecting Mn 2� directly in to the brain, we cantrace neuronal pathways using standard T1-weighted MRI.

To locate the injection site, we recorded neuronal activity within themBF before the injection and verified that the neurons there were Type 2neurons. We used a custom-made injectrode consisting of an epoxy-coated tungsten microelectrode for neuron recording and a silica tube forMn 2� injection. Before the injection, the injectrode was placed at theregion of the highest concentration of Type 2 neurons and the monkeyperformed the appetitive/aversive procedure (Fig. 1A). We injected 0.2�l of 150 mM solution of manganese chloride in to the Type 2 hotspotwithin the mBF. This concentration was previously shown to be effectiveand nontoxic (Simmons et al., 2008; Eschenko et al., 2010). We made theinjection in to the mBF over the duration of �20 min and waited for 15min to retract. After this, we retracted the injectrode 0.5 mm and waitedfor 15 min, after which we retracted the injectrode from the brain andprepared the monkey for MRI.

Figure 4. Properties of Type 2 neurons in mBF. A, Activity of 39 Type 2 neurons classified by PCA analysis (Fig. 2) during 100%, 50%, and 0% predictions of rewards and punishments. For eachneuron, the activity was normalized by dividing the trial firing rate by the baseline firing rate. B, Proportion of Type 2 neurons (left column) and other mBF neurons (right column) displaying aversivesignals (row 1), airpuff response (activity after airpuff � greater than activity before airpuff; row 2), and reward magnitude signals (row 3). Proportion of neurons displaying all three types of signals(rows 1–3) is shown in row 4. The proportion of Type 2 neurons displaying all three activity biases (64%) was significantly higher than the random combination (random combination � 12.5%binomial test; p � 0.0.01). C–F, Baseline firing characteristics of all Type 2 neurons recorded during all experiments in this study compared with other neurons in the mBF. Bar plots represent means.Single red dots indicate single neuron data. C, Firing irregularity indices. D, Autocorrelation width. E, Spike wave trough-to-trough duration. F, Baseline firing rate. *p � 0.05 (Wilcoxon rank sumtest). ns, Not significant.

Monosov et al. • Neurons in the Primate Medial Basal Forebrain J. Neurosci., May 13, 2015 • 35(19):7443–7459 • 7447

MR anatomical images were acquired in a 4.7 T horizontal scanner(Bruker Biospec 47/40) using a modified driven equilibrium Fouriertransform (MDEFT). The monkey’s head was placed in the scanner instereotactic position. A single loop circular coil was place on top of theanimal’s head. To minimize changes in RF across scans, we attached theMR surface coil directly to the monkey’s head holder (in the head im-plant). The whole-brain MDEFT images were acquired in a 3D volumewith a field of view 96 � 96 � 70 mm 3, and 0.5 mm isotropic voxel size.The read-out had an 11 ms repetition time, a 4.1 ms echo time, and a 11.6degree flip angle. The MDEFT preparation had a 1240 ms preinversiontime, and a 960 ms postinversion time for optimized T1 contrast at 4.7 T.Each 3D volume took 25.5 min to acquire without averaging, and 51 minif average by two.

Three baseline preinjection scans were collected a week earlier. Postin-jection scans were collected 24, 48, and 96 h after the injection. All scanswere obtained using the averaged 51 min acquisition. Those scans were

performed under isoflurane gas anesthesia, and 3–5 scans were collectedin each session. On the injection day, three 25 min scans were collected1 h after the injection was finished under ketamine anesthesia to localizethe injection site.

As in previous monkey MEMRI experiments, we could not detectmanganese transport by eye 24 h after the injection (Simmons et al.,2008). To visualize the transport, we calculated the percentage increase ofvoxel intensity after the injection by comparing averaged preinjectionscan with averaged postinjection scans. Before doing this, the averagedMRIs were processed by the Analysis of Functional NeuroImages (AFNI)toolkit (Cox, 1996). First, the image nonuniformity was reduced (AFNIfunction: 3dUniformize). Second, the images were resampled, doublingthe number of voxels (AFNI function: 3dresample). Third, the postinjec-tion scans were aligned with the average preinjection baseline scan usingaffine transforms (AFNI function: 3dAllineate). Fourth, the alignment ofpostinjection scans was further refined using a nonlinear warping algo-

Figure 5. Response of single mBF Type 2 neurons to rewards and punishments. A, Activity of single Type 2 neurons during 100%, 50%, and 0% predictions of rewards and punishments (n � 14:5 from Monkey H, 9 from Monkey P). For each neuron, the activity was normalized by dividing the trial firing rate by the baseline firing rate. B, Their average activity in the appetitive block.Conventions are the same as in Figure 1. C, Left, Difference in activity during trials in which rewards were delivered and not delivered after the 50% CS epoch (B, red vs pink trace). Single neuronactivity (black) and average (blue) are shown. Right, Baseline-subtracted activity of the same neurons to unexpected rewards during the ITI. *p � 0.05, significant difference between the twoconditions (left vs right; paired signed rank test). D, Average activity in the aversive block. E, Left, Difference in activity during trials in which airpuffs were delivered and not delivered after the 50%CS epoch. Right, Baseline-subtracted activity to unexpected airpuffs during the ITI. Same conventions as in C. Inset, Difference between less predicable airpuffs (after 50% CS epoch) and fullypredictable airpuff responses (after 100% CS epoch).

7448 • J. Neurosci., May 13, 2015 • 35(19):7443–7459 Monosov et al. • Neurons in the Primate Medial Basal Forebrain

rithm (AFNI function: 3dQwarp). Fifth, the averaged images (pre andpost) were smoothed with a 3D Gaussian filter (� � 0.2 mm). Last, beforecalculating the percentage increase, each average scan was normalized bydividing by the average intensity value of 5 mm 3 of cortex.

ResultsSeptal-basal forebrain complex contains distinct populationsof neurons sensitive to reward uncertainty (Experiment 1)We defined the mBF as a region containing the medial septumand the nucleus of the diagonal band of Broca. We recordedactivity of single neurons in the mBF and surrounding septalregions while monkeys participated in a Pavlovian procedure inwhich appetitive and aversive blocks were alternated (Fig. 1A). Inthe appetitive block, three visual CSs predicted juice with 100%,50%, and 0% chance. In the aversive block, three CSs predictedairpuffs with 100%, 50%, and 0% chance. Four monkeys (Mon-keys P, S, Sm, and T) participated in this procedure.

We found a group of neurons that showed reward uncer-tainty-selective activity. An example is shown in Figure 1B. In theappetitive block, the neuron displayed a monotonic increase in

firing rate during the uncertain (50% juice) CS period. The ac-tivity was truncated in response to the outcome (i.e., delivery oromission of juice). In contrast, the neuron showed no increase inactivity during the certain (100% and 0% juice) CS periods. Theneuron was also sensitive to the uncertain CS in the aversiveblock, but not selectively. It displayed similar ramping activityduring the uncertain (50% airpuff) and certain (100% airpuff)CS periods. The neuron also showed a phasic response to thedelivery of airpuff.

Among 370 neurons recorded in the septal-mBF complex, 70neurons showed reward uncertainty-selective activity. To quan-titatively test the heterogeneity among the reward uncertaintyneurons, we performed principal component analysis (PCA) us-ing the magnitudes of the neuron’s responses to the 6 CSs in theappetitive and aversive procedure (Fig. 2). The PCA analysisshowed that the reward uncertainty neurons were categorizedinto two distinct types: Type 1 (n � 31) and Type 2 (n � 39). Type1 and 2 neurons were different in several aspects: anatomicallocations within the septal-basal forebrain complex, the pattern

Figure 6. Differential sensitivities of mBF Type 2 neurons to reward amount and probability. A, Average activity of 31 mBF Type 2 neurons (15 in Monkey H, 16 in Monkey P) in a reward probabilityblock (top) and a reward amount block (bottom). In the reward amount block, five fractal CSs indicated five different amounts of juice. In the reward probability block, five other fractal CSs indicatedfive different probabilities of juice. Expected values of the CSs in the two blocks were matched. TS, Trial-start cue; US, juice delivery. B, Average normalized CS responses of the same neurons forreward probability and reward amount CSs. Right, CS responses of single neurons: probability (black), amount (gray); normalized to the maximum CS response, from 0 to 1. C, Probability-amountdifference in the activity of the mBF Type 2 neurons. The subtraction was done between a probability CS and an amount CS with an equal value. D, Normalized probability-amount CS responses. *p �0.05 (paired sign-rank test). ns, Not significant. Error bars indicate SE.

Monosov et al. • Neurons in the Primate Medial Basal Forebrain J. Neurosci., May 13, 2015 • 35(19):7443–7459 • 7449

Figure 7. Monkeys’ choice preference between CSs associated with different reward amounts and different reward probabilities. The monkey made a choice between two CSs amongthe 10 well-learned CSs (5 indicating reward amounts, 5 indicating reward probabilities) (Fig. 6). A, Choice between two reward probability CSs (left) and between two reward amountCSs (right). Each number indicates the choice percentage of the higher value CS. B, Choice percentage of a single reward probability CS versus all the other reward probability CSs (red).Choice percentage of a single reward amount CS versus all the other reward amount CSs (black). Data are compiled from a dataset of 3417 choice trials performed by Monkeys H and P.

Figure 8. Sensitivity of Type 2 neurons to reward variance. A, Experimental design. Four fractal CSs indicated four different levels of reward variance. Possible reward amounts are indicated withasterisks above each fractal. The large and small reward amounts occurred equally likely. The expected value was the same across the CSs. S.D, SD of reward amount. B, Average activity of 29 mBFType 2 neurons (16 in Monkey H, 13 in Monkey P), shown separately for each of the four fractals. C, CS responses shown individually for the 29 neurons (normalized to the maximum CS response; from0 to 1). D, Average CS responses of the same neurons. *p � 0.05 (paired sign-rank test). Error bars indicate SE.

7450 • J. Neurosci., May 13, 2015 • 35(19):7443–7459 Monosov et al. • Neurons in the Primate Medial Basal Forebrain

of CS and US responses, background firing rates, and other elec-trophysiological features (Fig. 3). The example neuron (Fig. 1B)belonged to Type 2 population. Notably, we found no punishmentuncertainty neurons (i.e., respond more strongly to 50% airpuff CSthan 100% or 0% airpuff CS) in the septal-mBF complex.

We found that Type 1 and 2 neurons were located largelyseparately in the septal-mBF complex (Fig. 3). Most Type 1 neu-rons (yellow dots) were located in the anterodorsal septum(ADS). They correspond to a group of reward uncertainty neu-rons reported previously (Monosov and Hikosaka, 2013). In con-trast, most Type 2 neurons (red dots) were located in the mBF(medial septum and diagonal band of Broca). The locations ofType 2 neurons were confirmed histologically (Fig. 3C). A total offive Type 2 neurons were recorded at and between the markinglesions (arrows), which were made along two electrode tracks.They showed very similar activity to the neuron shown in Figure1B. All of them were located in the diagonal band of Broca. In therest of the manuscript, we will focus on Type 2 neurons located inthe mBF.

Characterization of Type 2 neurons in the mBFUnlike Type 1 neurons in the ADS, single Type 2 neurons in themBF (hereafter called “mBF Type 2 neurons”) consistently dis-played responses to information other than reward uncertainty.These signals were combined in a rather unique way. This isshown qualitatively in Figure 4A in which activity of 39 Type 2neurons clustered by PCA analysis (Fig. 2) is superimposed. First,in addition to the ramping activity selective for reward uncer-tainty (50% juice CS), they showed similar ramping activity whenairpuff was expected (100% or 50%; the slopes of the ramping forreward uncertainty and airpuff expectation were correlatedacross neurons; � � 0.69; p � 0.01). Second, they were phasicallyactivated by airpuff. Third, in the appetitive block, their activitytended to be higher during 100% CS than 0% CS. A majority ofreward uncertainty-sensitive neurons in the mBF (Type 2)showed the same types of activity bias (Fig. 4A,B). In particular,the proportion of neurons that combined the three activity biases(64%) was significantly higher than the random combination(random combination � 12.5% binomial test; p � 0.0.01). Thiscombination of activity was not present among mBF neuronsthat were not sensitive to reward uncertainty (n � 211) (Fig. 4B).

These data suggest that the reward uncertainty-sensitive neuronsin the mBF comprise a group of neurons that were characterizedby the unique combination of activity.

Based on that observation, we wondered whether Type 2 neu-rons’ intrinsic firing properties set them apart from other mBFneurons. We found some differences (Fig. 4C–E). First, Type 2 neu-rons had lower firing irregularity indices than other mBF neurons(Fig. 4C; p � 0.05) and had lower variance of ISIs than other mBFneurons (p � 0.05). Second, Type 2 neurons’ firing was lessbursty than other mBF neurons, which was shown by two mea-surements (Viskontas et al., 2007): autocorrelation width (Fig.4D; p � 0.05) and ISI ratio (p � 0.05; Materials and Methods).Third, Type 2 neurons had narrower spike durations than othermBF neurons (Fig. 4E; p � 0.05). Fourth, however, there was nosignificant difference in their firing rate (Fig. 4F).

mBF Type 2 neurons signal unexpected outcomes(Experiment 2)In this and following experiments and analyses, we examinedhow mBF Type 2 neurons signal uncertainty and value, and howtheir responses change during learning. As shown before, mBFType 2 neurons showed strong predictive activity before an un-certain outcome of reward but were insensitive to the outcome(Fig. 2D, bottom).

We found, however, that mBF Type 2 neurons respondedstrongly when the reward was delivered unexpectedly outside thetask context (Fig. 5). In a separate set of experiment using thesame Pavlovian procedure (Fig. 1A), we delivered juice or airpuff(with the same amount and intensity) occasionally and randomlyduring the ITIs. mBF Type 2 neurons responded phasically toboth juice and airpuff (Fig. 5C,E, right). The results indicate thatmBF Type 2 neurons signal a reward outcome only when it isdelivered unexpectedly outside the behavioral context in whichthe monkey is currently engaged. The context dependency wasless selective for aversive outcomes.

mBF Type 2 neurons combine reward uncertainty and value(Experiment 3)While the neural selectivity for reward uncertainty was the mostpronounced aspect of Type 2 neural activity, there was an addi-tional asymmetry in the responses to the two certain conditions.

Figure 9. Monkeys’ choice preference between CSs associated with different reward variances. The monkey made a choice between two CSs among the four well-learned CSs, indicating differentreward variances (Fig. 8). A, Choice between two reward variance CSs. B, Choice percentage of a single reward variance CS versus all the other reward variance CSs. Data are compiled from a datasetof 4672 choice trials performed by Monkeys H and P.

Monosov et al. • Neurons in the Primate Medial Basal Forebrain J. Neurosci., May 13, 2015 • 35(19):7443–7459 • 7451

As shown in Figure 2D (bottom), the certain reward condition(100%) gave rise to a larger response than the certain no-rewardcondition (0%) (p � 0.05).

To investigate this further, we recorded Type 2 activity using aPavlovian procedure that contained two distinct contexts: a re-ward probability block and a reward amount block (Fig. 6). In thereward probability block, five CSs indicated five different proba-bilities (0%, 25%, 50%, 75%, 100%) of 0.4 ml of juice. In thereward amount block, five other CSs indicated five differentamounts of juice (0, 0.1, 0.2, 0.3, 0.4 ml). Thus, the expectedvalues (calculated as follows: amount � probability) of the CSswere matched between the two blocks.

In the reward amount block, the activity of mBF Type 2 neu-rons during the CS period varied with reward values, which wererepresented by reward amounts (Fig. 6A, bottom). Their activityincreased nonlinearly as the value increased but nearly plateauedat the two highest levels (Fig. 6B, gray). Thus, in the absence ofreward uncertainty, the activity of these neurons was consistentlymodulated by expected value of the CS.

However, in the reward probability block, where expectedvalue and reward uncertainty covaried, activity reflected a com-bination of these variables. In anticipation of the reward, neuralfiring rate increased as the expected value of acquiring reward

increased, but then decreased at the highest levels where the un-certainty went to zero (Fig. 6B, black). Overall, the activity ofmBF Type 2 neurons was higher in the reward probability blockthan in the reward amount block, specifically when the CS indi-cated reward uncertainty. As in Experiment 1, the anticipatoryincreases in activity were mostly observed in response to uncer-tain reward predictions, not changes in certain reward.

The data thus far suggest that BF Type 2 neurons encode thecombination of value and uncertainty. If so, the difference inactivity between the probability and amount blocks would showtheir sensitivity to reward uncertainty. Indeed, as shown in Figure6C, the subtracted activity grew larger monotonically during theCS period only when reward was uncertain. It was highest whenthe reward probability was 50% than 25% or 75%. Overall, thesubtracted activity during the CS period showed an inverseU-shaped pattern (Fig. 6D), matching previous findings of Type1 neurons in the ADS (Monosov and Hikosaka, 2013). This com-bination of results raises the intriguing possibility that BF Type 2neurons’ activity reflects a specific convergence of neural inputs.Type 2 neurons may be the recipient of uncertainty-selective sig-nal from ADS Type 1 neurons as well as a value-selective signalfrom other neurons located elsewhere in the brain. A straightfor-ward additive combination of these two signals would account

Figure 10. Changing activity of Type 2 neurons during uncertainty resolution. Left and center columns, As Monkey P experienced new sets of fractal CSs on the appetitive task (Fig. 1A, left) across3 d (one session each day), the activity pattern of mBF Type 2 neurons changed nonuniformly in time. Before the neuronal recording, the monkey had experienced multiple sets of fractal CSs withdifferent degrees (e.g., 0, 1, or 2 d), which enabled the comparison of the same mBF neuron’s activity across different degrees of learning. Right column, The monkey also had experienced a set offractal CSs with many daily sessions (�100), which revealed the same neuron’s responses to the well-learned CSs.

7452 • J. Neurosci., May 13, 2015 • 35(19):7443–7459 Monosov et al. • Neurons in the Primate Medial Basal Forebrain

for the complex properties observed in theType 2 neurons, particularly because thesubtracted activity (Fig. 6C) showed nohint of a value signal (Fig. 6A, bottom).

The stronger responses of BF Type 2neurons to uncertain CSs might be relatedto the monkey’s preference of reward un-certainty. However, this possibility wasnot supported by the following experi-ment: We presented two CSs simultane-ously and let the monkey choose one. Themonkey’s choice was largely guided by theexpected reward value in any of the fol-lowing conditions: when the choice wasmade between two reward probabilityCSs or between two reward-size CSs(Fig. 7).

Alternatively, the stronger responses ofBF Type 2 neurons to uncertain CSs mightbe caused by their response selectivity tovisual objects. To test this possibility, weused two sets of CSs for 11 of the 31 Type2 neurons recorded in Experiment 3. Wefound that their responses were similar forboth sets (p � 0.1), suggesting that theresponses of BF Type 2 neurons were sen-sitive to reward uncertainty, not visual ob-ject identity.

mBF Type 2 neurons signal rewardvariance (Experiment 4)In the probability block of Experiment 3,changes in uncertainty were always asso-ciated with changes in value because thereward amount was held constant. If un-certainty and value signals in BF Type 2neurons have distinct origins, then thoseneurons should encode the quantitativelevel of uncertainty when the value is heldconstant.

We tested this hypothesis using a Pav-lovian procedure in which we varied thevariance of reward amount while the ex-pected value remained unchanged acrossall the trials (Fig. 8A). Among four CSsused, one was a certain CS and the otherthree were uncertain CSs. Each of the un-certain CSs predicted two amounts ofjuice with an equal probability (i.e., 50%).Across the CSs, the average of the twoamounts remained the same, but the dif-ference between the two amounts varied.

Figure 11. Time courses of neuronal and behavioral changes during uncertainty resolution. A, For each of the first, second, andthird day of training (Day 1–3), average responses of Type 2 neurons to 100%, 50%, and 0% CSs are shown at the start and end oftraining (left and center columns). Their responses to well-learned CSs are shown for comparison (right column). *p�0.05. ns, Notsignificant. Red square represents the first time when data for the newly learned CSs became statistically not different from data forthe well-learned CSs. B, Behavioral changes during choice trials: choice probability of the higher valued CSs (top), target acquisitiontime during the choice (middle), and the number of saccades before choice (bottom). The behavioral data are shown for choicebetween 100% and 50% CSs, given the slower changes in neuronal responses between 100% and 50% CSs (A). Error bars indicateSE. C, Neuronal learning expressed as changes in differential responses between the uncertain forced single stimulus (50%) CS

4

trials and the 100% CS trials during Day 1 (first learning ses-sion) quantified by ROC (ROCs � 0.5 indicate uncertainty se-lectivity). Type 2 neuron selectivity is plotted in black. Forcontrast, we recorded additional Type 1 ADS neurons duringthe same learning task (shown in red; 3 in Monkey Sm and5 in Monkey P). As shown in A, Type 2 neurons learnedslowly, whereas Type 1 neurons quickly displayed uncer-tainty preference as shown previously (Monosov and Hiko-saka, 2013, their Fig. 5).

Monosov et al. • Neurons in the Primate Medial Basal Forebrain J. Neurosci., May 13, 2015 • 35(19):7443–7459 • 7453

The difference (or variance) is often taken as a measure of risk(Weber et al., 2004; Preuschoff and Bossaerts, 2007; Platt andHuettel, 2008; Bach and Dolan, 2012).

We found that BF Type 2 neurons encoded reward variance(or risk) (Fig. 8B–D). Their average activity increased monoton-ically during the trial-start cue and CS period until the resolutionof uncertainty (Fig. 8B). Importantly, the increase in activity dur-ing the CS period was higher when the reward variance washigher (Fig. 8D). This tendency was common among 29 BF neu-rons tested (Fig. 8C).

Unlike Experiment 3, all the CSs were associated with thesame expected reward value, but with different levels of rewardvariance. When we presented two of the four CSs (Fig. 8A) simul-taneously, the monkeys tended to choose the higher-variance(i.e., riskier) CS (Fig. 9). The result confirmed previous findings(McCoy and Platt, 2005; Platt and Huettel, 2008; O’Neill andSchultz, 2010; So and Stuphorn, 2010).

mBF Type 2 neurons slowly acquire reward uncertainty signal(Experiment 5)Experiments 1– 4 were performed after the monkeys had learnedthe meanings of all CSs: the associated values and the degree ofuncertainty. However, when the monkeys experienced the CSsfor the first time, the meaning of each CS was unknown. Howdo BF Type 2 neurons acquire their sensitivity to value anduncertainty?

To answer this question, we tested the same type of neurons(i.e., Type 2) across daily sessions using CSs with different degreesof learning (Fig. 10). The task was the same as used in Experiment1 but included only the appetitive block. In each session, we usedthree sets of objects as CSs: (1) a new set of objects, (2) a notwell-learned set of objects (learned once or twice), and (3) awell-learned set of objects (learned � 1 month).

In response to new CSs (Fig. 10, Day 1, left), these Type 2neurons increased their activity monotonically for all CSs until

Figure 12. Transport of manganese chloride to the HF. A, The location of the injection site is visualized as a bright spot on sagittal (two images on the left) and coronal (right) MR images. Type2 neurons recorded in the same monkey (classified by PCA analysis in Fig. 2) are shown in red (A, left). The 95% confidence ellipse around the neurons is shown in yellow. The same image is also shownwithout the neurons to observe the injection site within the confidence ellipse. White lines indicate 5 mm. CD, Caudate nucleus; PUT, putamen; vmPF, ventromedial prefrontal cortex. B, Percentageincrease of voxel intensity (“MEMRI labeling”) shown in a parasagittal plane 24 h (B) and 96 h (C) after the manganese chloride injection. aHF, Anterior HF; pHF, posterior HF. D–F, MEMRI labelingshown in a coronal plane 24 h (E) and 96 h (F) after the injection. The coronal plane is focused on the HF whose structure is shown in D. CA fields are indicated within the hippocampus. TF,Parahippocampal cortical area TF. B, White line indicates 10 mm. D, White line indicates 5 mm. B, Red line indicates the location of the coronal plane shown in D–F.

7454 • J. Neurosci., May 13, 2015 • 35(19):7443–7459 Monosov et al. • Neurons in the Primate Medial Basal Forebrain

the delivery of an outcome (i.e., juice or no juice). Their re-sponses changed by the end of the 1 day learning session (Fig.10, Day 1, middle): the 0% CS elicited a strong suppression( p � 0.05). Thus, after 1 d learning, the response to the 0% CSreached the level of the well-learned response (Fig. 11A, bot-tom row). In contrast, their responses to the 100% and 50% CSremained unchanged (Figs. 10, 11A, top and middle rows).This response pattern was similar to their value-coding (Fig.6A, bottom), except that their activity ramped up toward theUS for all CSs.

It took �2 days for the neurons to develop their uncer-tainty selectivity (Fig. 10, middle row). This occurred as the BFType 2 neurons decreased their responses to 100% CS, so thatthe responses to 50% CS became larger than both the re-sponses to 0% CS and 100% CS ( p � 0.05; paired signed ranktest). By the end of the 2 day learning session, their responsesto all CSs were not significantly different from their responsesto well-learned CSs (Fig. 11A, center column). However, theresponses to 100% CS appeared less suppressed even after the2 day or 3 day learning session than after the long-term learn-ing. In summary, the slow time course of uncertainty selectiv-ity was due to a gradual decrease in the magnitude of the 100%CS responses.

How might the changes in the activity of BF Type 2 neuronsbe related to the monkey’s behavior? To investigate this issue,after every 12 single CS trials, the monkey experienced ninechoice trials. During these trials, the monkey chose one CSamong two by fixating it for 800 ms. The monkey was free tolook around at the CSs before making his choice for up to 5 s.The details of this procedure are explained in Materials andMethods.

By the end of the 1 day learning session, the monkeys alreadychose the higher-valued CS on most trials, as they do for thewell-learned CSs (Fig. 11B, left column, top). Correspondingly,BF Type 2 neurons appeared to have acquired value signals by theend of the 1 day learning session (see above). With further learn-ing, however, the monkey’s behavior continued to change: deci-sion speed became faster. This was shown as a continued decreasein the target acquisition time and the number of saccades madebefore the target acquisition (Fig. 11B, middle and bottomrows). Only by the end of the 2 day learning session, thesevalues became indistinguishable from those for well-learnedCSs. These data suggest that: (1) long-term learning is re-quired for the ramping activity of BF Type 2 neurons to betuned to the uncertainty of reward outcome; and (2) this pro-cess might be related to the monkey’s confidence or skill tochoose high-valued objects.

It is noteworthy that the reward uncertainty activity devel-oped at different speeds between Type 2 neurons in the mBF andType 1 neurons in the ADS (Fig. 11C). Unlike Type 2 neurons,Type 1 neurons developed uncertainty selectivity within one ses-sion on the first day of learning, as we previously showed (Mono-sov and Hikosaka, 2013). The data provide another distinctfeature between Type 1 and Type 2 neurons.

Type 2 hotspot projects to the intermediate HF (Experiment 6)The function of mBF Type 2 neurons depends on where theirsignals are sent via anatomical projections. To address this issuefor the population of mBF neurons we recorded, we usedMEMRI. As manganese is taken up and transported antero-gradely by neurons and is also visible in MRI scans, this methodallows for in vivo tracing of neuronal connections in monkeys(Saleem et al., 2002; Simmons et al., 2008) and then allows the

subject to continue to participate in tasks/experiments afterthe tracing procedure is completed. Using electrophysiologi-cally guided positioning of an injection cannula, we intro-duced manganese chloride solution (MnCl2, 0.2 �l, 150 mM)into the mBF area in which we found many Type 2 neurons(Fig. 12A). This can be qualitatively observed by comparingFigure 12A with Figure 3. To quantitatively verify this obser-vation, we also generated a 95% confidence ellipse in MRIspace from the locations of the Type 2 neurons recorded in thesame monkey (Fig. 12A) and confirmed that the injection waswithin this ellipse (Fig. 12B).

We subsequently acquired a series of high-resolution MRIscans across a period of �100 h to visualize how manganese wastransported (Materials and Methods).

Twenty-four hours after the injection, we found MEMRIlabeling most prominently in the intermediate region of theipsilateral HF (Fig. 12 B, E), including the hippocampus andmedial temporal cortices. The signal continued to increaseover the course of days (Fig. 12C,F ). The contralateral HF wasalso labeled similarly but more weakly (data not shown). Noobvious labeling was found in other brain areas. These exper-iments suggest the possibility that the signals of mBF Type 2neurons are transmitted to the HF, especially in its intermedi-ate region.

DiscussionWithin the mBF in the macaque monkey, a group of neuronsencoding reward uncertainty (Type 2 neurons) also signaled anumber of other internal variables related to motivation and ex-pectancy (Fig. 4B). These responses occurred in two contexts:before or after an outcome. First, the preoutcome responses oc-curred when the monkey was engaged in a task, reflecting themonkey’s prediction. The reward uncertainty response appearedas a monotonic increase in activity that was rapidly truncatedafter the outcome (reward or no reward). A similar monotonicincrease of activity occurred when the monkey anticipated a cer-tain or uncertain punishment. The reward value responseappeared as a change in tonic firing (Fig. 6). Second, the pos-toutcome responses occurred strongly when an outcome wasdelivered unexpectedly. The unexpected nature was particu-larly clear for the postreward response (Fig. 5). In contrast, nopostreward response occurred during task performance, evenwhen the reward was uncertain (Fig. 2D, bottom), a situationthat creates reward prediction error (Schultz et al., 1997; Ma-tsumoto and Hikosaka, 2009; Bromberg-Martin et al., 2010).Therefore, BF Type 2 neurons encode “surprise” rewards (orstrong/highly salient outcomes), but not reward predictionerror. These results indicate that single mBF Type 2 neuronsencoded a unique combination of information: reward uncer-tainty, expected reward value, punishment prediction, sur-prise rewards and punishments.

Our experiments suggest that the different kinds of informa-tion encoded by BF Type 2 neurons may originate from differentsources. For example, BF Type 2 neurons combined informationabout “value” (or reward size) and information about uncer-tainty additively (Fig. 6). Also, value selectivity developed earlierduring learning than uncertainty selectivity (Fig. 10). The distinctnature of value and uncertainty signals in BF Type 2 neurons mayprovide their target areas with a signal about behavioral relevancethat is balanced between value and uncertainty and partly basedon the subject’s experience.

A likely source of the reward uncertainty signal in BF Type 2neurons is the ADS where we found neurons that selectively en-

Monosov et al. • Neurons in the Primate Medial Basal Forebrain J. Neurosci., May 13, 2015 • 35(19):7443–7459 • 7455

coded reward uncertainty (Monosov and Hikosaka, 2013). In thisreport, we classified them as Type 1 neurons, with their locationsshown in Figure 3 and their activity shown in Figure 2 (togetherwith Type 2 neurons). Notably, the inverse U-shaped uncertaintycoding of BF Type 2 neurons (Fig. 6D) was very similar to that ofADS Type 1 neurons (Monosov and Hikosaka, 2013) (Fig. 3C).The connection from ADS Type 1 neurons to BF Type 2 neuronsmay be supported by some rodent anatomical studies (Meibachand Siegel, 1977; Russchen et al., 1985; Staiger and Nurnberger,1991). Interestingly, Type 1 neurons developed the reward un-certainty response quickly in one learning session, whereas Type2 neurons acquired reward uncertainty selectivity by eliminat-ing their responses to certain CSs and did so slowly across days(Fig. 10). Therefore, it is unlikely that the reward uncertaintysignal is directly transmitted from Type 1 to Type 2 neurons.Instead, a connection might act as a conditioning signal forType 2 neurons such that their activity during the uncertainCS (not certain CSs) was preserved. In other words, the ADScould be, in effect, training the mBF to shape its responsesbased on uncertainty.

On the other hand, the sources of the other signals in BF Type2 neurons are unknown. One candidate may be the medial tha-lamic nuclei (Hsu and Price, 2009), which receive dense inputsfrom the ventromedial prefrontal cortex (Hsu and Price, 2007,2009) where distinct populations of neurons tonically encodeappetitive or aversive states (Monosov and Hikosaka, 2012).Other candidates include the brainstem serotonergic and adren-ergic nuclei (Russchen et al., 1985; Semba et al., 1988; Vertes,1988; Espana and Berridge, 2006).

What then is the function of mBF Type 2 neurons? Lesion andbrain imaging studies have linked the mBF to learning and mem-ory functions (Damasio et al., 1985; Ridley et al., 1988, 1989,1999a; Morris et al., 1992; Roman et al., 1993; Abe et al., 1998; DeRosa and Hasselmo, 2000; Fujii et al., 2002; De Rosa et al., 2004;Baxter and Bucci, 2013; Iglesias et al., 2013). Anatomically, themBF is shown to project to brain areas that are heavily involved inlearning and memory, including the HF and the anterior cingu-late cortex (Mesulam et al., 1983). We thus consider the possibil-ity that mBF Type 2 neurons serve learning and/or memoryfunctions.

A fundamental question is as follows: Why do mBF Type 2neurons encode a relatively fixed combination of informationrelated to both rewards and punishments? Interestingly, singlemBF Type 2 neurons encoded the combination of the reward-related and punishment-related information in a unidirectionalmanner: they increased their firing rate in reward-associated andpunishment-associated contexts (Fig. 2D, bottom). Therefore,Type 2 signals would not be suitable to guide future actionsaiming at obtaining rewards and avoiding punishments (e.g.,through reinforcement learning). Consistent with Type 2 neu-rons not being a source of reinforcing signals are two addi-tional observations: (1) Type 2 neurons did not encode rewardprediction errors; and (2) they responded more strongly to50% than 100% rewards. These data indicate that Type 2 neu-rons could not reliably guide behavior toward rewards in un-certain contexts.

On the other hand, the same combination of informationwould be suitable for creating or modifying episodic memories aswell as for other monitoring functions (e.g., attention for learn-ing). For instance, we remember past events more likely if theywere associated with something rewarding, punishing, or sur-prising (Loewenstein et al., 2001; Dolcos et al., 2004a, b; Hulse etal., 2007). Consistent with this hypothesis, MEMRI tracing

showed that the region of the mBF where Type 2 neurons wereconcentrated projected to the HF, which is a key region for form-ing and recalling episodic or abstract associative memories (Has-selmo et al., 1996; Mishkin et al., 1997; Vargha-Khadem et al.,1997; Dore et al., 1998; Wallenstein et al., 1998; Manns et al.,2003; Ergorul and Eichenbaum, 2004). The HF is known to re-ceive continuous information about the external world from ce-rebral cortical areas and the internal state from neuromodulatorysystems, including the dopaminergic system (Lisman and Grace,2005). The signals from the mBF could instruct the HF to take a“snapshot” of the ongoing events and contexts when they arerewarding, punishing, or surprising. These “snapshots” of theworld would be used as important evidence for the animal’s judg-ment and decision making (Tulving, 2002; Eichenbaum, 2004;Bar, 2009; Squire and Wixted, 2011; Stokes et al., 2012).

Computational models have suggested that the mBF exertscontrol over learning and episodic memory in the HF (Has-selmo et al., 1995, 1996; Myers et al., 1996). In support of thesemodels and our hypothesis, disruptions of the HF and mBFcan produce similar deficits in memory formation (Ridley etal., 1988, 1989, 1999a; Morris et al., 1992; Roman et al., 1993;Suzuki et al., 1993, 1997; Abe et al., 1998; Turchi et al., 2005,2008).

More experiments are now necessary to further elucidatethe role of mBF Type 2 neurons in cognitive functions. Somestudies have suggested that, in certain contexts, the HF con-tributes to the reduced attention to irrelevant events ratherthan the increased attention to relevant events (Han et al.,1995), which is different from what we proposed above (i.e.,attention for learning). Notably, this alternative function ofthe HF may be guided by cholinergic neurons in the mBF(Chiba et al., 1995; Baxter et al., 1997; Oros et al., 2014). Onthe other hand, noncholinergic mBF neurons may be respon-sible for increasing learning and memory capacities (Dwyer etal., 2007; Pang et al., 2011; Baxter and Bucci, 2013; Roland etal., 2014). This consideration suggests that Type 2 neurons inthe mBF are noncholinergic. Indeed, unlike Type 2 neurons(Fig. 4C–F ), cholinergic neurons typically have low firingrates, wide spike durations, and display irregular firing(Manns et al., 2000a, b; Momiyama and Zaborszky, 2006; Si-mon et al., 2006; Hedrick and Waters, 2010; Unal et al., 2012).However, the neurochemical profile of Type 2 neurons is un-known. Although BF is often characterized by concentratedgroups of cholinergic neurons (Mesulam et al., 1983; Za-borszky et al., 2008), it contains other kinds of neurons, in-cluding GABAergic and glutamatergic neurons (Semba, 2000;Lin et al., 2006; Zaborszky et al., 2008), all of which mayproject to the HF (Everitt and Robbins, 1997; Semba, 2000;Easton et al., 2012). Therefore, finding out whether Type 2neurons belong to a particular neuronal group would be im-portant to reveal the circuit mechanisms that control what toremember and from what to learn.

ReferencesAbe K, Inokawa M, Kashiwagi A, Yanagihara T (1998) Amnesia after a dis-

crete basal forebrain lesion. J Neurol Neurosurg Psychiatry 65:126 –130.CrossRef Medline

Anderson BA (2013) A value-driven mechanism of attentional selection.J Vis 13:3. CrossRef Medline

Bach DR, Dolan RJ (2012) Knowing how much you don’t know: a neuralorganization of uncertainty estimates. Nat Rev Neurosci 13:572–586.CrossRef Medline

Bar M (2009) The proactive brain: memory for predictions. Philos Trans RSoc Lond B Biol Sci 364:1235–1243. CrossRef Medline

7456 • J. Neurosci., May 13, 2015 • 35(19):7443–7459 Monosov et al. • Neurons in the Primate Medial Basal Forebrain

Baxter MG, Bucci DJ (2013) Selective immunotoxic lesions of basal fore-brain cholinergic neurons: twenty years of research and new directions.Behav Neurosci 127:611– 618. CrossRef Medline

Baxter MG, Holland PC, Gallagher M (1997) Disruption of decrements inconditioned stimulus processing by selective removal of hippocampalcholinergic input. J Neurosci 17:5230 –5236. Medline

Berry SD, Thompson RF (1979) Medial septal lesions retard classical condi-tioning of the nicitating membrane response in rabbits. Science 205:209 –211. CrossRef Medline

Bromberg-Martin ES, Matsumoto M, Hikosaka O (2010) Dopamine in mo-tivational control: rewarding, aversive, and alerting. Neuron 68:815– 834.CrossRef Medline

Chiba AA, Bucci DJ, Holland PC, Gallagher M (1995) Basal forebrain cho-linergic lesions disrupt increments but not decrements in conditionedstimulus processing. J Neurosci 15:7315–7322. Medline

Courville AC, Daw ND, Touretzky DS (2006) Bayesian theories of condi-tioning in a changing world. Trends Cogn Sci 10:294 –300. CrossRefMedline

Cox RW (1996) AFNI: software for analysis and visualization of functionalmagnetic resonance neuroimages. Comput Biomed Res 29:162–173.CrossRef Medline

Damasio AR, Graff-Radford NR, Eslinger PJ, Damasio H, Kassell N (1985)Amnesia following basal forebrain lesions. Arch Neurol 42:263–271.CrossRef Medline

Davies RM, Gerstein GL, Baker SN (2006) Measurement of time-dependentchanges in the irregularity of neural spiking. J Neurophysiol 96:906 –918.CrossRef Medline

Daye PM, Monosov IE, Hikosaka O, Leopold DA, Optican LM (2013) py-Electrode: an open-source tool using structural MRI for electrode posi-tioning and neuron mapping. J Neurosci Methods 213:123–131. CrossRefMedline

De Rosa E, Hasselmo ME (2000) Muscarinic cholinergic neuromodula-tion reduces proactive interference between stored odor memoriesduring associative learning in rats. Behav Neurosci 114:32– 41.CrossRef Medline

De Rosa E, Desmond JE, Anderson AK, Pfefferbaum A, Sullivan EV (2004)The human basal forebrain integrates the old and the new. Neuron 41:825– 837. CrossRef Medline

Dolcos F, LaBar KS, Cabeza R (2004a) Interaction between the amygdalaand the medial temporal lobe memory system predicts better memory foremotional events. Neuron 42:855– 863. CrossRef Medline

Dolcos F, LaBar KS, Cabeza R (2004b) Dissociable effects of arousal andvalence on prefrontal activity indexing emotional evaluation and subse-quent memory: an event-related fMRI study. Neuroimage 23:64 –74.CrossRef Medline

Dore FY, Thornton JA, White NM, Murray EA (1998) Selective hippocam-pal lesions yield nonspatial memory impairments in rhesus monkeys.Hippocampus 8:323–329. CrossRef Medline

Dwyer TA, Servatius RJ, Pang KC (2007) Noncholinergic lesions of the me-dial septum impair sequential learning of different spatial locations.J Neurosci 27:299 –303. CrossRef Medline

Easton A, Douchamps V, Eacott M, Lever C (2012) A specific role for sep-tohippocampal acetylcholine in memory? Neuropsychologia 50:3156 –3168. CrossRef Medline

Eccleston C, Crombez G (1999) Pain demands attention: a cognitive-affective model of the interruptive function of pain. Psychol Bull 125:356 –366. CrossRef Medline

Eichenbaum H (2004) Hippocampus: cognitive processes and neural repre-sentations that underlie declarative memory. Neuron 44:109 –120.CrossRef Medline

Ergorul C, Eichenbaum H (2004) The hippocampus and memory for“what,” “where,” and “when.” Learn Mem 11:397– 405. CrossRef Medline

Esber GR, Haselgrove M (2011) Reconciling the influence of predictivenessand uncertainty on stimulus salience: a model of attention in associativelearning. Proc Biol Sci 278:2553–2561. CrossRef Medline

Eschenko O, Canals S, Simanova I, Logothetis NK (2010) Behavioral, elec-trophysiological and histopathological consequences of systemic manga-nese administration in MEMRI. Magn Reson Imaging 28:1165–1174.CrossRef Medline

Espana RA, Berridge CW (2006) Organization of noradrenergic efferents toarousal-related basal forebrain structures. J Comp Neurol 496:668 – 683.CrossRef Medline

Everitt BJ, Robbins TW (1997) Central cholinergic systems and cognition.Annu Rev Psychol 48:649 – 684. CrossRef Medline

Everitt BJ, Sirkia TE, Roberts AC, Jones GH, Robbins TW (1988) Distribu-tion and some projections of cholinergic neurons in the brain of thecommon marmoset, Callithrix jacchus. J Comp Neurol 271:533–558.CrossRef Medline

Fujii T, Okuda J, Tsukiura T, Ohtake H, Miura R, Fukatsu R, Suzuki K,Kawashima R, Itoh M, Fukuda H, Yamadori A (2002) The role of thebasal forebrain in episodic memory retrieval: a positron emission tomog-raphy study. Neuroimage 15:501–508. CrossRef Medline

Han JS, Gallagher M, Holland P (1995) Hippocampal lesions disrupt decre-ments but not increments in conditioned stimulus processing. J Neurosci15:7323–7329. Medline

Hasselmo ME, Schnell E, Barkai E (1995) Dynamics of learning and recall atexcitatory recurrent synapses and cholinergic modulation in rat hip-pocampal region CA3. J Neurosci 15:5249 –5262. Medline

Hasselmo ME, Wyble BP, Wallenstein GV (1996) Encoding and retrieval ofepisodic memories: role of cholinergic and GABAergic modulation in thehippocampus. Hippocampus 6:693–708. CrossRef Medline

Hedrick T, Waters J (2010) Physiological properties of cholinergic and non-cholinergic magnocellular neurons in acute slices from adult mouse nu-cleus basalis. PLoS One 5:e11046. CrossRef Medline

Hsu DT, Price JL (2007) Midline and intralaminar thalamic connectionswith the orbital and medial prefrontal networks in macaque monkeys.J Comp Neurol 504:89 –111. CrossRef Medline

Hsu DT, Price JL (2009) Paraventricular thalamic nucleus: subcortical con-nections and innervation by serotonin, orexin, and corticotropin-releasing hormone in macaque monkeys. J Comp Neurol 512:825– 848.CrossRef Medline

Hulse LM, Allan K, Memon A, Read JD (2007) Emotional arousal andmemory: a test of the poststimulus processing hypothesis. J Psychol 120:73–90. Medline

Iglesias S, Mathys C, Brodersen KH, Kasper L, Piccirelli M, den Ouden HE,Stephan KE (2013) Hierarchical prediction errors in midbrain and basalforebrain during sensory learning. Neuron 80:519 –530. CrossRefMedline

Kesner RP (1988) Reevaluation of the contribution of the basal forebraincholinergic system to memory. Neurobiol Aging 9:609 – 616. CrossRefMedline

Le Pelley ME, Beesley T, Griffiths O (2011) Overt attention and predictive-ness in human contingency learning. J Exp Psychol Anim Behav Process37:220 –229. CrossRef Medline

Lin SC, Gervasoni D, Nicolelis MA (2006) Fast modulation of prefrontalcortex activity by basal forebrain noncholinergic neuronal ensembles.J Neurophysiol 96:3209 –3219. CrossRef Medline

Lisman JE, Grace AA (2005) The hippocampal-VTA loop: controlling theentry of information into long-term memory. Neuron 46:703–713.CrossRef Medline

Loewenstein GF, Weber EU, Hsee CK, Welch N (2001) Risk as feelings.Psychol Bull 127:267–286. CrossRef Medline

Manns ID, Alonso A, Jones BE (2000a) Discharge properties of juxtacellu-larly labeled and immunohistochemically identified cholinergic basalforebrain neurons recorded in association with the electroencephalogramin anesthetized rats. J Neurosci 20:1505–1518. Medline

Manns ID, Alonso A, Jones BE (2000b) Discharge profiles of juxtacellularlylabeled and immunohistochemically identified GABAergic basal fore-brain neurons recorded in association with the electroencephalogram inanesthetized rats. J Neurosci 20:9252–9263. Medline

Manns JR, Hopkins RO, Squire LR (2003) Semantic memory and the hu-man hippocampus. Neuron 38:127–133. CrossRef Medline

Matsumoto M, Hikosaka O (2009) Two types of dopamine neuron dis-tinctly convey positive and negative motivational signals. Nature 459:837– 841. CrossRef Medline

McCoy AN, Platt ML (2005) Risk-sensitive neurons in macaque posteriorcingulate cortex. Nat Neurosci 8:1220 –1227. CrossRef Medline

McNally GP, Johansen JP, Blair HT (2011) Placing prediction into the fearcircuit. Trends Neurosci 34:283–292. CrossRef Medline

Meibach RC, Siegel A (1977) Efferent connections of the septal area in therat: an analysis utilizing retrograde and anterograde transport methods.Brain Res 119:1–20. CrossRef Medline

Mesulam MM, Mufson EJ, Levey AI, Wainer BH (1983) Cholinergic inner-vation of cortex by the basal forebrain: cytochemistry and cortical con-

Monosov et al. • Neurons in the Primate Medial Basal Forebrain J. Neurosci., May 13, 2015 • 35(19):7443–7459 • 7457

nections of the septal area, diagonal band nuclei, nucleus basalis(substantia innominata), and hypothalamus in the rhesus monkey.J Comp Neurol 214:170 –197. CrossRef Medline

Mishkin M, Suzuki WA, Gadian DG, Vargha-Khadem F (1997) Hierarchi-cal organization of cognitive memory. Philos Trans R Soc Lond B Biol Sci352:1461–1467. CrossRef Medline

Momiyama T, Zaborszky L (2006) Somatostatin presynaptically inhibitsboth GABA and glutamate release onto rat basal forebrain cholinergicneurons. J Neurophysiol 96:686 – 694. CrossRef Medline

Monosov IE, Hikosaka O (2012) Regionally distinct processing of rewardsand punishments by the primate ventromedial prefrontal cortex. J Neu-rosci 32:10318 –10330. CrossRef Medline

Monosov IE, Hikosaka O (2013) Selective and graded coding of reward un-certainty by neurons in the primate anterodorsal septal region. Nat Neu-rosci 16:756 –762. CrossRef Medline

Moore DJ, Keogh E, Eccleston C (2012) The interruptive effect of pain onattention. Q J Exp Psychol 65:565–586. CrossRef Medline

Morris MK, Bowers D, Chatterjee A, Heilman KM (1992) Amnesia follow-ing a discrete basal forebrain lesion. Brain 115:1827–1847. CrossRefMedline

Muir JL, Page KJ, Sirinathsinghji DJ, Robbins TW, Everitt BJ (1993) Ex-citotoxic lesions of basal forebrain cholinergic neurons: effects onlearning, memory and attention. Behav Brain Res 57:123–131.CrossRef Medline

Myers CE, Ermita BR, Harris K, Hasselmo M, Solomon P, Gluck MA (1996)A computational model of cholinergic disruption of septohippocampalactivity in classical eyeblink conditioning. Neurobiol Learn Mem 66:51–66. CrossRef Medline

Nakamura K, Matsumoto M, Hikosaka O (2008) Reward-dependent mod-ulation of neuronal activity in the primate dorsal raphe nucleus. J Neuro-sci 28:5331–5343. CrossRef Medline

Ogawa M, van der Meer MA, Esber GR, Cerri DH, Stalnaker TA, SchoenbaumG (2013) Risk-responsive orbitofrontal neurons track acquired salience.Neuron 77:251–258. CrossRef Medline

O’Neill M, Schultz W (2010) Coding of reward risk by orbitofrontal neu-rons is mostly distinct from coding of reward value. Neuron 68:789 – 800.CrossRef Medline

Oros N, Chiba AA, Nitz DA, Krichmar JL (2014) Learning to ignore: amodeling study of a decremental cholinergic pathway and its influenceon attention and learning. Learn Mem 21:105–118. CrossRef Medline

Pang KC, Jiao X, Sinha S, Beck KD, Servatius RJ (2011) Damage of GABAergicneurons in the medial septum impairs spatial working memory and extinc-tion of active avoidance: effects on proactive interference. Hippocampus 21:835–846. CrossRef Medline

Pautler RG, Silva AC, Koretsky AP (1998) In vivo neuronal tract tracingusing manganese-enhanced magnetic resonance imaging. Magn ResonMed 40:740 –748. CrossRef Medline

Pearce JM, Hall G (1980) A model for Pavlovian learning: variations in theeffectiveness of conditioned but not of unconditioned stimuli. PsycholRev 87:532–552. CrossRef Medline

Platt ML, Huettel SA (2008) Risky business: the neuroeconomics of de-cision making under uncertainty. Nat Neurosci 11:398 – 403. CrossRefMedline

Preuschoff K, Bossaerts P (2007) Adding prediction risk to the theory ofreward learning. Ann N Y Acad Sci 1104:135–146. CrossRef Medline

Ridley RM, Baker HF, Murray TK (1988) Basal nucleus lesions in monkeys:recognition memory impairment or visual agnosia? Psychopharmacology95:289 –290. Medline

Ridley RM, Aitken DM, Baker HF (1989) Learning about rules but notabout reward is impaired following lesions of the cholinergic projec-tion to the hippocampus. Brain Res 502:306 –318. CrossRef Medline

Ridley RM, Pugh P, Maclean CJ, Baker HF (1999a) Severe learning impair-ment caused by combined immunotoxic lesion of the cholinergic projec-tions to the cortex and hippocampus in monkeys. Brain Res 836:120 –138.CrossRef Medline

Ridley RM, Barefoot HC, Maclean CJ, Pugh P, Baker HF (1999b) Differenteffects on learning ability after injection of the cholinergic immunotoxinME20.4IgG-saporin into the diagonal band of Broca, basal nucleus ofMeynert, or both in monkeys. Behav Neurosci 113:303–315. CrossRefMedline

Roland JJ, Stewart AL, Janke KL, Gielow MR, Kostek JA, Savage LM,Servatius RJ, Pang KC (2014) Medial septum-diagonal band of Broca

(MSDB) GABAergic regulation of hippocampal acetylcholine efflux isdependent on cognitive demands. J Neurosci 34:506 –514. CrossRefMedline

Roman FS, Simonetto I, Soumireu-Mourat B (1993) Learning and mem-ory of odor-reward association: selective impairment following hori-zontal diagonal band lesions. Behav Neurosci 107:72– 81. CrossRefMedline

Russchen FT, Amaral DG, Price JL (1985) The afferent connections of thesubstantia innominata in the monkey, Macaca fascicularis. J Comp Neurol242:1–27. CrossRef Medline

Saleem KS, Pauls JM, Augath M, Trinath T, Prause BA, Hashikawa T, Logo-thetis NK (2002) Magnetic resonance imaging of neuronal connectionsin the macaque monkey. Neuron 34:685–700. CrossRef Medline

Savage LM, Roland J, Klintsova A (2007) Selective septohippocampal- butnot forebrain amygdalar- cholinergic dysfunction in diencephalic amne-sia. Brain Res 1139:210 –219. CrossRef Medline

Schultz W, Dayan P, Montague PR (1997) A neural substrate of predictionand reward. Science 275:1593–1599. CrossRef Medline

Semba K (2000) Multiple output pathways of the basal forebrain: organiza-tion, chemical heterogeneity, and roles in vigilance. Behav Brain Res 115:117–141. CrossRef Medline

Semba K, Reiner PB, McGeer EG, Fibiger HC (1988) Brainstem afferents tothe magnocellular basal forebrain studied by axonal transport, immuno-histochemistry, and electrophysiology in the rat. J Comp Neurol 267:433–453. CrossRef Medline

Simmons JM, Saad ZS, Lizak MJ, Ortiz M, Koretsky AP, Richmond BJ (2008)Mapping prefrontal circuits in vivo with manganese-enhanced magneticresonance imaging in monkeys. J Neurosci 28:7637–7647. CrossRefMedline

Simon AP, Poindessous-Jazat F, Dutar P, Epelbaum J, Bassant MH (2006)Firing properties of anatomically identified neurons in the medial septumof anesthetized and unanesthetized restrained rats. J Neurosci 26:9038 –9046. CrossRef Medline

So NY, Stuphorn V (2010) Supplementary eye field encodes option andaction value for saccades with variable reward. J Neurophysiol 104:2634 –2653. CrossRef Medline

Solca M, Di Pietro M, Schnider A, Leemann B (2015) Impairment of seman-tic memory after basal forebrain and fornix lesion. Neurocase 21:198 –205. CrossRef Medline

Squire LR, Wixted JT (2011) The cognitive neuroscience of human memorysince H.M. Annu Rev Neurosci 34:259 –288. CrossRef Medline

Staiger JF, Nurnberger F (1991) The efferent connections of the lateral sep-tal nucleus in the guinea pig: intrinsic connectivity of the septum andprojections to other telencephalic areas. Cell Tissue Res 264:415– 426.CrossRef Medline

Stokes MG, Atherton K, Patai EZ, Nobre AC (2012) Long-term memoryprepares neural activity for perception. Proc Natl Acad Sci U S A 109:E360 –E367. CrossRef Medline

Suzuki WA, Zola-Morgan S, Squire LR, Amaral DG (1993) Lesions of theperirhinal and parahippocampal cortices in the monkey produce long-lasting memory impairment in the visual and tactual modalities. J Neu-rosci 13:2430 –2451. Medline

Suzuki WA, Miller EK, Desimone R (1997) Object and place memory in themacaque entorhinal cortex. J Neurophysiol 78:1062–1081. Medline

Tulving E (2002) Episodic memory: from mind to brain. Annu Rev Psychol53:1–25. CrossRef Medline

Turchi J, Saunders RC, Mishkin M (2005) Effects of cholinergic deafferen-tation of the rhinal cortex on visual recognition memory in monkeys.Proc Natl Acad Sci U S A 102:2158 –2161. CrossRef Medline

Turchi J, Buffalari D, Mishkin M (2008) Double dissociation of pharmaco-logically induced deficits in visual recognition and visual discriminationlearning. Learn Mem 15:565–568. CrossRef Medline

Unal CT, Golowasch JP, Zaborszky L (2012) Adult mouse basal forebrainharbors two distinct cholinergic populations defined by their electrophys-iology. Front Behav Neurosci 6:21. CrossRef Medline

Vargha-Khadem F, Gadian DG, Watkins KE, Connelly A, Van Paesschen W,Mishkin M (1997) Differential effects of early hippocampal pathologyon episodic and semantic memory. Science 277:376 –380. CrossRefMedline

Vertes RP (1988) Brainstem afferents to the basal forebrain in the rat. Neu-roscience 24:907–935. CrossRef Medline

Viskontas IV, Ekstrom AD, Wilson CL, Fried I (2007) Characterizing interneu-

7458 • J. Neurosci., May 13, 2015 • 35(19):7443–7459 Monosov et al. • Neurons in the Primate Medial Basal Forebrain

ron and pyramidal cells in the human medial temporal lobe in vivo usingextracellular recordings. Hippocampus 17:49–57. CrossRef Medline

Voytko ML (1996) Cognitive functions of the basal forebrain cholinergicsystem in monkeys: memory or attention? Behav Brain Res 75:13–25.CrossRef Medline

Wallenstein GV, Eichenbaum H, Hasselmo ME (1998) The hippocampus asan associator of discontiguous events. Trends Neurosci 21:317–323.CrossRef Medline

Weber EU, Shafir S, Blais AR (2004) Predicting risk sensitivity in humansand lower animals: risk as variance or coefficient of variation. Psychol Rev111:430 – 445. CrossRef Medline

Whitehouse PJ, Price DL, Struble RG, Clark AW, Coyle JT, Delon MR (1982)Alzheimer’s disease and senile dementia: loss of neurons in the basalforebrain. Science 215:1237–1239. CrossRef Medline

Wiech K, Tracey I (2013) Pain, decisions, and actions: a motivational per-spective. Front Neurosci 7:46. CrossRef Medline

Yu AJ, Dayan P (2005) Uncertainty, neuromodulation, and attention. Neu-ron 46:681– 692. CrossRef Medline

Zaborszky L, Hoemke L, Mohlberg H, Schleicher A, Amunts K, Zilles K(2008) Stereotaxic probabilistic maps of the magnocellular cellgroups in human basal forebrain. Neuroimage 42:1127–1141.CrossRef Medline

Zaborszky L, Csordas A, Mosca K, Kim J, Gielow MR, Vadasz C, NadasdyZ (2015) Neurons in the basal forebrain project to the cortex in acomplex topographic organization that reflects corticocortical con-nectivity patterns: an experimental study based on retrogradetracing and 3D reconstruction. Cereb Cortex 25:118 –137. CrossRefMedline

Monosov et al. • Neurons in the Primate Medial Basal Forebrain J. Neurosci., May 13, 2015 • 35(19):7443–7459 • 7459